library(data.table)
library(dplyr)
library(dplyr)
library(tidyr)
library(plotly)
library(lubridate)In this homework you should use plotly unless said otherwise.
To create pdf version of your homework, knit it first to html and then print it to pdf. Interactive plotly plots can be difficult sometimes to convert to static images suitable for insertion to LaTex documents (that is knitting to PDF).
Look for questions in R-chunks as comments and plain text (they are
prefixed as Q
“The Iris flower data set or Fisher’s Iris data set is a multivariate data set introduced by the British statistician and biologist Ronald Fisher in his 1936 paper The use of multiple measurements in taxonomic problems as an example of linear discriminant analysis” https://en.wikipedia.org/wiki/Iris_flower_data_set
# Q1.1. Read the iris.csv file (2 points)
# hint: use fread from data.table, it is significantly faster than default methods
# be sure to have strings as factors (see stringsAsFactors argument)
df <- fread("iris.csv",stringsAsFactors=TRUE)# Q1.2. Show some values from data frame (2 points)
head(df,6)# Q1.3. Build histogram plot for Sepal.Length variable for each species using plot_ly
# (use color argument for grouping) (2 points)
# should be one plot
plot_ly(df, x = ~Sepal.Length, color = ~Species, colors = "Dark2", type="histogram")# Q1.4. Repeat previous plot with ggplot2 and convert it to plotly with ggplotly (2 points)
p <- ggplot(df, aes(x = Sepal.Length)) + geom_histogram(bins = 8,position = "dodge",aes(fill = Species))
ggplotly(p)# Q1.5. Create facet 2 by 2 plot with histograms similar to previous but for each metric
# (2 points)
# hint:
# following conversion to long format can be useful:
# iris %>% gather(key = "metric", value = "value",-Species)
#
iris_gathered <- gather(df, metric, value, -Species)
iris_gatherediris_facet <- ggplot(iris_gathered, aes(value, fill = Species))+ geom_histogram(position = "dodge") +facet_wrap(~ metric)
ggplotly(iris_facet)## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.Q1.6. Which metrics has best species separations? (2 points)
Petal Length and Petal Width are the best for species separation as they segregate the data best.
# Q1.7. Repeat above plot but using box plot (2 points)
plot_ly(data = iris_gathered, x = ~value, y = ~interaction(Species,metric), color = ~Species, type = "box")# Q1.8. Choose two metrics which separates species the most and use it to make scatter plot
# color points by species (2 points)
plot_ly(data = df, x = ~Petal.Length, y = ~Petal.Width, color = ~Species, type = "scatter")## No scatter mode specifed:
## Setting the mode to markers
## Read more about this attribute -> https://plotly.com/r/reference/#scatter-mode# Q1.9. Choose three metrics which separates species the most and use it to make 3d plot
# color points by species (2 points)
plot_ly(data = df, x = ~Petal.Length, y = ~Petal.Width, z= ~Sepal.Length, color = ~Species)## No trace type specified:
## Based on info supplied, a 'scatter3d' trace seems appropriate.
## Read more about this trace type -> https://plotly.com/r/reference/#scatter3d## No scatter3d mode specifed:
## Setting the mode to markers
## Read more about this attribute -> https://plotly.com/r/reference/#scatter-modeQ1.10. Comment on species separation (2 points): The feature that seperates the species best are Petal Length and Petal Width. Among the three species, Setosa is the most seperated based, while Virginica & versicolor are pretty similar in some edge cases. Virginica has the Longest Sepal & Petal & thickest Petals meanwhile Setosa is short and all the three parameters. Versicolor lies in the middle and all these three parameters.
Download us-states.csv (there is also a copy in homework assignment) from https://github.com/nytimes/covid-19-data/. README.md for details on file content.
# Q2.1. Read us-states.csv (2 points)
us <- fread("us-states.csv",stringsAsFactors=TRUE)# Q2.2. Show some values from dataframe
head(us,6)# Q2.3. Create new dataframe with new cases per month for each state (2 points)
# hint:
# is cases column cumulative or not cumulative?
us$year <- year(ymd(us$date))
us$month <- month(ymd(us$date))
us$day <- day(ymd(us$date))
df$date <- as.POSIXct( df$date, format="%Y/%m/%d" )
us_grp_state <- us %>% group_by(state,year,month) %>%
summarise(date = max(date),fips = max(fips),cases = max(cases), deaths = max(deaths))## `summarise()` has grouped output by 'state', 'year'. You can override using the
## `.groups` argument.final_us_state <- us_grp_state %>% group_by(state) %>% mutate(lag = shift(cases, 1, 0, type="lag"), cases, total_cases_month = (cases - lag) )
final_us_state# Q2.4.Using previous dataframe plot new monthly cases in states, group by states
# The resulting plot is busy, use interactive plotly capabilities to limit number
# of displayed states
# (2 points)
plot_ly(data = final_us_state, x = ~date , y = ~total_cases_month , color = ~state, type = 'scatter', mode = "line")%>%
layout(xaxis = list(title = 'year_month'),
yaxis = list(title = 'cases')
)## Warning in RColorBrewer::brewer.pal(N, "Set2"): n too large, allowed maximum for palette Set2 is 8
## Returning the palette you asked for with that many colors
## Warning in RColorBrewer::brewer.pal(N, "Set2"): n too large, allowed maximum for palette Set2 is 8
## Returning the palette you asked for with that many colors# Q2.5.Plot new monthly cases only in NY state
# (2 points)
new_york_state <- final_us_state %>% filter(state == "New York")
plot_ly(data = new_york_state , x = ~date , y = ~total_cases_month, type = "scatter")%>%
layout(xaxis = list(title = 'year_month'),
yaxis = list(title = 'cases')
)## No scatter mode specifed:
## Setting the mode to markers
## Read more about this attribute -> https://plotly.com/r/reference/#scatter-mode# Q2.6. Found the year-month with highest cases in NY state
# (2 points)
max_cases_new_york <- new_york_state %>% group_by(state) %>% slice(which.max(total_cases_month))
max_cases_new_york_final <- max_cases_new_york %>% select(-year,-month,-lag)
max_cases_new_york_final# Q2.7. Plot new cases in determined above year-month
# using USA state map, color each state by number of cases (2 points)
# hint:
# there two build in constants in R: state.abb and state.name
# to convert full name to abbreviation
us_state_jan_2022 <- final_us_state %>% filter( year == 2022,month == 1)
us_state_jan_2022 <- us_state_jan_2022 %>% mutate(code = state.abb[match(state,state.name)])
g <- list(
scope = 'usa',
projection = list(type = 'albers usa'),
lakecolor = toRGB('white')
)
plot_geo() %>%
add_trace(
z = ~us_state_jan_2022$total_cases_month, text=us_state_jan_2022$state, span = I(0),
locations=us_state_jan_2022$code, locationmode = 'USA-states'
) %>%
layout(geo = g) %>%
colorbar(title = "cases")# Q2.8. Add animation capability (2 points)
# hint:
# for variable frame you need either integer or character/factorial so
# convert date to character or factorial
#final_us_state <- final_us_state %>% mutate(date = format(date, "%Y-%m"))
anim <- final_us_state %>% mutate(date = format(as.character((date))), code = state.abb[match(state,state.name)])
g <- list(
scope = 'usa',
projection = list(type = 'albers usa'),
lakecolor = toRGB('white')
)
animplot_geo() %>%
add_trace(
z = ~anim$total_cases_month, text=anim$state, span = I(0),
locations=anim$code, locationmode = 'USA-states',frame = ~anim$date
) %>%
layout(geo = g) %>%
colorbar(title = "cases")%>%
animation_slider(
currentvalue = list(prefix = "date: ")
)Q2.9. Compare animated plot from Q2.8 to plots from Q2.4/Q2.5 (When you would prefer one or another?) (2 points)
The animated plot is more user friendly and provides better access to the time series data hence I would prefer it. The plot in Q2.4 is useful for studying the pattern across USA. It will be useful if we want to study a pattern in USA but otherwise in a general scenario I would prefer the animated graph. The plot in Q2.5 is limited and shows only one state, New York. It will be useful if we want to study a pattern only in New York but otherwise in a general scenario I would prefer the animated graph.